Grammars in a broad sense (specifications of structural commitments) are complex artefacts that define software languages. Assessing and improving their quality in an automated, non-idiosyncratic manner is an unsolved problem which we face in an especially acute form in the case of mass maintenance of hundreds of heterogeneous grammars (parser specs, ADTs, metamodels, XML schemata, etc) in the Grammar Zoo. In an attempt to apply software language engineering methods to solve a software language

more »

... engineering problem, we design a language for grammar mutations capable of applying uniform intentional transformations in the scope of a big grammar or a corpus of grammars. In this paper, we describe a disciplined process of engineering such a language by systematic reuse of semantic components of another existing software language. The constructs of the reference language are analysed and classified by their intent, each category of constructs is then subjected to rewriting. This process results in a set of constructs that form the new language.